US3355719A - Analog voltage memory circuit - Google Patents

Description

NOV. 28, 1967 5 R FOX ANALOG VOLTAGE MEMORY CIRCUIT 2 Sheets-Sheet l Filed Oct.

Nov. 28, 1967 s. R. Fox

ANALOG VOLTAGE MEMORY CIRCUIT 2 Sheets-Sheet 2 Filed Oct.

INVENTOR. STEPHEN RICHARD FOX ATTORNEY United States Patent O-ice 3,355,7l9 Patented Nov.. 28, 1967 3,355,719 ANALOG VOLTAGE MEMORY ClRCUlT Stephen Richard Fox, Los Angeles, Salif., assigner, by mesne assignments, to the United States of America as represented by the Secretary of the Navy Filed Oct. 8, 1963, Ser. No. 314,823 Claims. (Cl. Mtl-173) The present invention relates to computers, and more particularly to an electrical network for incorporation therein which is capable of either (1) storing an analog voltage for subsequent utilization, or (2) receiving a plurality of voltage increments and developing an output representing the algebraic sum thereof.

Certain computer applications require that the information received by the computer be remembered substantially without change for an indefinite period of time. One example is found in the launching of missiles from submarines, where the various steps in orienting each missile in firing position are programmed by a computing device which must take into account a number of varying factors such as the speed of the vessel, its direction of movement, and the sea and air -conditions existing at the time the operation is carried out.

Any system which is capable of satisfying the above requirements must necessarily be of at least moderate complexity. In one arrangement which has been developed, means are provided whereby a number of guidance capsules (for missiles of the Polaris type) are provided with complete launching controls by a single computing unit which incorporates as many as forty-eight energy-storage components each of which is adapted to receive varying amounts of electrical energy, the particular amount received by any one component representing a desired missile control function.

It will be recognized that in such a system the data being stored must remain essentially unchanged in magnitude regardless of the length of time that may elapse before such data is utilized. In computers where storage is effected by a series of binary memory elements the problem is not so severe, but where electrostatic components, such as capacitors, are made use of, then it is a fairly common occurrence for the charge on the capacitor to be bled olf through any distributed circuit impedances, or to gradually leak off to ground. When this takes place, the output energy from the capacitor is no longer an accurate representation of the input data, and consequently quite serious errors may arise.

It is a feature of the present invention to provide means whereby an energy lost from a capacitor in the above manner is automatically replenished and thereby precluded from dropping below a predetermined level. This is brought about by operation of a servomechanism which acts to feed back a prescribed amount of energy the magnitude of which is determined by the length of time that the servomechanisms feedback is operative. This time period, in turn, is governed by the amount of energy that has been lost by the storage component in question.

One important characteristic of the present concept is that the storage device is adapted to receive input data in the form of incremental bits over a period of time. Within a range represented by each individual data bit a certain stable voltage point is established, and the servomechanism acts to maintain the stored voltage at this stable point regardless of the amount of energy that may subsequently be lost thereby, either through leakage or in the distributed circuit impedances.

In order to accomplish the above objective, it is only necessary that two regularly-recurring reference waves be available, one of such waves being a series of narrow pulses at a given repetition frequency, and the other being a preferably square wave at a frequency which is an integral multiple of the pulse frequency. The input data, on the other hand, may either be an analog voltage to be remembered, or a series of sequentially-received pulses the algebraic sum of which is desired.

One object of the present invention, therefore, is to provide an improved circuit for storing information until some subsequent time when it is to be utilized.

Another object of the invention is to provide an ener-gystorage network in which any received information is maintained at a constant amplitude level regardless of forces which may tend to diminish the amount of energy so stored.

A further object of the invention is to provide an energy-storage network in which means are present for automatically replenishing energy which is lost from the network between the time of its reception and the time of utilization.

Other objects, advantages, and novel features of the invention will become apparent from the following detailed description of the invention when considered in conjunction with the accompanying drawings wherein;

FIG. 1 is a schematic diagram of a circuit intentionally simplified in order to illustrate the principles of operation of the present invention;

FIG. 2 is a scale indicating the stable voltage points of the servomechanism;

FIG. 3 is a block diagram of one practical circuit embodying the principles of the invention;

FIG. 4 is a set of waveforms useful in explaining the operation of the circuit of FIG. 3; and

FIG. 5 is a circuit diagram setting forth the details of each of the individual networks included in FIG. 3.

The essential feature of the present concept is the provision of an energy-storage device to which an analog source voltage may be applied, this voltage having a value E lying in the range between E0 and -l-Eo. The energystorage device may take the form of a capacitor 10. Connected in parallel with capacitor 10 is a load 11 which continually bleeds the capacitor 10 of any charge which may reside thereon. In order to exactly replace the charge bled from the storage device, a voltage V is applied through an impedance 12 by the appropriate periodic closure of gates 14 and 16.

FIGURE 2 of the drawing illustrates the principle of operation of the disclosed system. As brought out thereby, the voltage range from no to -l-E is divided into n increments, each of which possesses one stable point indicated by a dot in the drawing. An important characteristie of the invention is that following the application of a particular analog volta-ge source to capacitor 10 which results in a change of the capacitors voltage, the subsequent simultaneous closings of both gates 14 and 16 will cause the capacitor to be driven to the nearest stable voltage when the driving source voltage is removed. Throughout the following description, the closing of a gate is intende-d to mean the establishment of a path for current flow. Also, if digital information is to be recorded, the application to the capacitor lil of any particular voltage increment whose magnitude is approximately AE will cause the servcmechanism to drive the capacitor to that stable point wlhin the range of the z'ncrement. The char-ge on the capacitor will be maintained at this stable point until the source voltage (digital or analog) is reapplied. If the source is analog, the capacitor 10 will again seek the potential of the source. If the source is digital, the capacitors voltage will be changed by some increment. Upon subsequent closing of both gates and removal of the driving source the voltage of the capacitor will be brought to the nearest stable point.

Mention has been made above of the electronic servomechanism which is Aeffective to bring the `char-ge `on the storage capacitor to the nearest stable point within a given ran-ge. A schematic showing of a preferred arrangement for .achieving this objective is set forth in FIG. 3. It will be recognized that the memory capacitor may be identical to tthe ycapacitor ot' FIG. l, -and .that the gates 14 .and 16 `may 'likewise be the same yin both figures. YIhe servomechanism further includes va Acurrent amplifier 20, to which the output of the memory capacitor 1.0` is applied, and a discharge network 22 connected in parallel with the series combination ofthe memory capacitor 10 and its current amplifier 20. In parallel with the discharge network 22 is the series Acombination of both gatesv 14 and 16 as well asa bias source 24. As shown in the drawings, the .-modied output of the current amplier 2t) represents the output of the system. At the same time, a portion 'of this .output lis applied to a so-called ramp generator .26, 'which also `receives a cyclically-carrying voltage from .a source 2-8. The output of source 28 is in the form .of a series of trigger pulses which Vwill be described in greater detail in connection with FIG. 4 of `the drawings. It is `only `important to .note at 'this point that the pulses applied to the ramp lgenerator 26 are lof a regularly-recurring nature and have a frequency fo. A pulse-forming .network 30 receives the output of the ramp generator 26.

r'Referring again lmomentarily to FIG. l=of the drawings, it will be recalled that the capacitor 1'0` only vreceives a charge from the voltage V when both gates 14 and 16 are closed to allow current to pass therethrough. To control the operation of these gates, a connection is made to gate 16 from the pulse-forming network 30` of FIG. 3, and `in addition a vwave of substantially square configuration is applied to gate 14 from .a source 32.

The :wave forms of FIG. 4 are believed .to'be helpful in .understanding the operation of the servomechanism of FlG. .3. The ramp generator 26 yis ytriggered Vby the pulses from generator 28 to produce a sawtooth wave 34 having a lfrequency `fn as shown in FIG. 4, while zthe square wave A36 produced lby source 32 has a period equal to where n yis a `who/le number. In the drawings, nis arbitrarily -chosen to `equal 3.

The memory :capacitor 10 of FIG. 1 sees 4va load -impedance 11 which continually bleeds ythe capacitor Aof any charge lwhich may be present thereon. 'If this capacitor is yto vretain or remember a voltage, then the net cur rent through it must be zero, and, consequently, any charge which has been bled oli through impedance -11 must be replaced, However, as `shown in FIG. 1, such replenishment can only occur when .both gates 14 and 16 are closed, at which time current can fliowinto the capacitor from the potential source.

The sawtooth 4wave 34, as illustrated in FIG. 4, has a frequency fo and a time period To. This wave 34 is initiated at the capacitor voltage level Ec and, as also shown in FIG. 4, decreases beyond .a predetermined .reference voltage level Er. The square wave 36, on the other hand, has la frequency njo and a time period wheren'is van integer greater than one.

The sawtooth -wave 34 constitutes the `output of .the ramp ,generator 26 of FIG. 3. This wave 34 is applied through `the pulse-forming network 30 to the gate 16 as .-a series of rectangular pulses 30 to maintan'the latter in closed circuit status for the duration of each pulse. Since current only flows into capacitor 10 when both the gates 14 and '16 are closed, and since the circuit status of gate 14 is-governed by thesquare wave 36, a simultaneous occurrence of a pulse 36 .and a pulse 38 is .necessary for activation of the servo loop. The time interval during which Vcurrent 'ow occurs Vis -represented by -the Ashaded port-ion of the control wave 36 (for gate 14) and of the control wave 36 (for gate 16). Expressed diiferently, the two gates are closed concurrently only during the shaded time interval in FIG. 4.

The operation of the circuit of FIG. 3 as a servomechanism will be clear .if it is .assumed (for example) that the time period during which .both gates 14 and 16 are concurrently closed is .too short, and that all of the charge bled otl the memory capacitor 1t) is not replaced. Under such circumstances, there is a net current flow out of the capacitor, and -its voltage yEc must decrease. Since Ec is of reduced value, the sawtooth wave 34 pro'-1 duced by ythe ramp generator 26 'must start from a lower voltage level (see FIG, 4) and hence vless .time will elapse before this wave 34 reaches the reference voltage level Er. Since the time instant .at which the wave 34 reaches this threshold level Er determines the point at which gate 16 closes (sor in other words the rising .edge of a control pulse 38 in FIG. 4) `-then the period during which -bozh gates 14'and 16 are closed is lengthened, `since the .repetition frequency of wave .36 is constant and `:its period remains unchanged. With the `time of simultaneous gate closing 'being thus lengthened, more current flows into `the memory capacitor from the potential source v24 to compensate for that `bled off through the .distributed circuit imped-ances. After a number of cycles of operation of the servomechauism in the manner above described, the voltage on the memory capacitor stabilizes itself so that `gate 16 closes at a time such that the currentcyolical ly .pulsed into the memory capacitor .exactly couuteracts the lcurrent bled oit" through .the circuit `impedances, .and the network .reaches a state of equilibrium at one of the points depicted .by the black dots in FIG. 2 of the drawings.

The above .discussion has demonstrated the phase-'lock nature of the servo loop which -includes the ramp igenerator 26 and the pulse-forming network :30. It should be noted, however, that the square-wave 36 lwhich is applied to .gate 14 to control the circuit 'status thereof has a frequency higher than the frequency of the sawtooth wave 34 yproduced by the ramp .generator v2.6, and hence of the pulses .38. In the example chosen, -the frequency of the .pulses 36 is three :times thattofthe pulses 38. Since gate 14 iscyclically closed three :times foreveryztime .that gate 1'6 is :closed '.by the yapplication thereto of :the pulses 38, there will ybe three stable voltages .at which -thepmemory capacitor 10 can remain. The precise -voltage vremembered thereby depends on whether -lock-ongis fachieved on the rst, second or ythird opening of .gate 14, or, in other words, on whether lock-on results during fthe first, second lor third Icycle of the square wave 36. Consequently, the -servo loop :has a theoretical `upper limit of accuracy equal .to

where :n equals the frequency of the wave 3'6 divided by the'frequency of the sawtoothfwave 34.

It will fnow be 'recognized that, since n discrete voltages can .bestoredlby the .fmemorycapacitor 10, `thecircuit illustrated in FIG. 3 can operate either as a .digital-to-analog converter or yas :a device -for :accumulating a series .of voltage increments the sum of which represents a desired output. For example, ifzthe ydeviceis-storinga voltage fsuch that lock-.on is effective on the .ich periodof gate 14, 'then a predeterminedquantity of :charge can be pulsed intothe memory capacitor 10 lsuchrthat lock-on is `achieved inthe v(--|-,1 )st period of .gate .14. Thus'lthe voltage on'the memory capacitor is stepped up and served to I-the vnext stable voltage that it permanently remembers (indicated by the black dots on FIG. 2) until another pulse of current drives -the capacitor either .(a) back down'so ythat lock-on again .occurs `on vthe ith interval, for r(b) further .up so that lock-on is effected -on the (i-l2)d interval. Hence,

both data accumulation and digital-to-analog conversion is accomplished.

In FIG. of the drawings is illustrated certain of the operating details of the system of FIG. 3. For example, the memory capacitor is shown as feeding into a highgain current amplifier which is designed to minimize loading. The output of this amplifier is at essentially the same level as the voltage on capacitor 10, and is applied to the ramp generator 26. The latter is triggered at a rate fo by pulses from generator 2S (FIG. 3). When the ramp generator output (the sawtooth wave 34) reaches the reference voltage level E,r (FIG. 4) the network 30 produces a pulse 38 which acts to close the gate 16. At the same time, gate 14 is being sequencially opened and closed by the square wave 36 from generator 32.

Considering now certain of the circuit details, it will be seen that the current amplier 20 consists of two cascaded emitter follower transistors 40 and 42. The voltage at the emitter of transistor 42 has a value two diode drops below the voltage on capacitor 10, and, consequently, is not exactly equal to the capacitor voltage Ec. In the ramp generator 26, a Zener diode #t4-impresses a constant voltage across the resistors between the emitter of a transistor 46 and -E2, making the collector of transistor 46 act as a constant-current sink. If a diode 48 in the pulse-forming network 30 is back-biased, this current viiow through transistor 46 also ilows through a capacitor 50 and, since the emitter of transistor 42 is held at two diode drops below EC, the voltage at point 52 decreases linearly with time. By applying a pulse of frequency fo to turn on transistor 54 and discharge capacitor 50, the negative-going sawtooth at point 52 is made to start from two diode drops below Ec. A

Referring again to the pulse-forming network 30, the emitter of transistor 56 is held at a negative potential (minus 7 volts, for example) by the Zener diode 58. If the voltage point 52 is much more positive than the emitter voltage of transistor 56, then diode 48 is back-biased, and the resistor 60 biases transistor 56 on. When transistor 56 is conducting, transistor 62 is cut off, and transistor 64 is on. As the sawtooth Voltage at point 52 decreases, diode 48 will start to conduct. This withdraws the bias that transistor 56 received from resistor 60. At this point, regeneration between transistors 56 and 62 quickly turns the former off, the latter on, and transistor 64 off. When transistor 64 goes olf, the two inductively-coupled coils 66 (one being part of the gate 16) will ring through capacitor 68 and the base-emitter diode portion of transistor 70, turning the latter on and closing gate 16.

It will be recalled that gate 14 is opened and closed at the rate of njo c.p.s. and, when both gates 14 and 16 are closed, current Hows into the capacitor 1t) due to the bias established by source E4 in the unit 24. In a representative circuit, typical values for fo and n are 1600 and 612, respectively. In the example given, njo has a value of 979.2 kc.

The discharge network 22 is designed so that a known quantity of current is bled from the memory capacitor 10. The circuit of FIGS. 3 and 5 functions such that T t fIcapucitor dtzj o Ibleed Irenlenish dt=0 where T0 is the period of the sawtoother wave 34 (FIG. 4) and t is the time during which both gates 14 and 16 are closed. Consequently, the better Ibleed is controlled, the easier it is to design the replenishing network.

It should be noted that both the discharge network 22 and the replenishing circuit iioat with the memory voltage, so that both Ibleed and Ireplemsh are not extensively dependent on the remembered voltage. n

The data input to capacitor 10 may comprise either (.1) an analog voltage the magnitude of which is to be remembered, or (2) a digital source to dump current pulses in-to the capacitor for accumulation and digital-to-analog conversion.

Obviously many modifications and variations of the present invention are possible in the light of the above teachings. It is therefore to be understood that within the scope of the appended claims the invention may be practiced otherwise than as specifically described.

I claim:

1. In a computing system wherein electrical energy representing data to be remembered is retained by a storage device under conditions where such energy is subject to gradual diminution over a period of time, such loss of energy resulting in the output data from the system no longer being an accurate representation of the data originally received thereby, the combination of:

a source of data to be remembered;

an energy-storage device receiving data from said source;

a second source of energy :to be supplied to said storage device to replenish energy gradually lost therefrom over a period of time;

means for controlling the amount of energy supplied to said storage device from said second energy source, said controlling means including rst and second gates connected in series relation;

means for selectively controlling the operation of said first and second gates so that, when both gates are concurrently activated to establish a current path therethrough, energy is supplied from said second source to said storage device to replenish energy which may have been gradually lost therefrom subsequent to the reception thereby of data from said source; and

means for regulating the operation of said last-mentioned means so that the time period when both said gates are concurrently activated to establish a current path therethrough is a function of the amount of energy present in said storage device.

2. A system for maintaining the amount of energy in a memory capacitor at a level which is representative of data previously received by said capacitor, said system including:

a memory capacitor adapted to receive data energy;

an external source of replenishment energy for said capacitor;

a circuit for controlling the ilow of replenishment energy into said capacitor to compensate for that lost thereby subsequent to the reception of said data, said circuit including a iirst source of gating pulses of a predetermined constant repetition frequency;

a first gate closed upon the reception thereby of a pulse from said rst source;

means for generating Ia cyclically-recurring control Wave the amplitude `of which is a function of the amount of data energy in said capacitor at any given instant of time;

a second source of gating pulses the time of occurrence of which is governed by the instantaneous amplitude of said control potential;

a second gate closed upon the reception thereby 4of a pulse from said second source; and

means for connecting said two gates in series with said external source of replenishment energy so that, upon 'the concurrent closing of both said gates following the reception by each gate of a pulse from its respective source, replenishment energy will flow into said capacitor to compensate for that lost subsequent to the reception of said data.

3. The combination of claim 2 in which said cyclicallyrecurring control wave is of substantially sawtooth conguration.

4. A system according to claim 2, in which said circuit for controlling the amount of replenishment energy which liows into said capacitor to compensate for that lost thereby subsequent to the reception of said data includes a generator developing an output wave of sawtooth conrfigurationythe instant of generationcf .each `cycleof said .sawtoothwavebeingdetermiried bythe level` of :the .energy .presentin=saidirnemorydevice at'suchitime, and means vfordeveloping an output :pulse duringeac'hzcycle whenever said sawtooth :wave :reaches apredeterminedtenergy level, the pulses so developed being applied to saidsecond gate ito -control fthe circuit status thereof.

'5. 'Ina computerincorporating an energysstorage .device -of thecapacitancel .type which-is adaptedto receive .energy :representative of data .to .be remembered, .and which -energy=is subject 5to gradual diminution over .a period vof 4time subsequentitofits reception bysaiddevice so that atte-r such period ottime :said energyLis nolonger representative of the data supplied thereto, the improvement whichcom- :prises .a circuit for maintaininga predetermined energy level on said capacitor which is determined bylthe'magniltude of :the s"toredfenergyat .any .given instant ,of time, said .circuit comprising:

a capacitor adapted to receive :energy-representative of .fdata :to "be remembered; a lcurrentamplifier in series .with said .capacitor and to whichtheoutput y.of said: capacitor is applied; a rst gate; .a second gate; ,a-:source of replenishment energy 'for said capacitor; tmeans for electricallyconnecting said tworgates and said source of replenishment energy in series with one another, vthe resulting `series combination `being con- .nected to said capacitor Aand yin vparallel with said current amplifier; .a..dischar.ge network connected in .parallel with .the series combination-of lsaid ...two gates and said sourceof retplenishmentlenergy, said discharge lnetworkacting to .diminish thecharge on said capacitor lat an essentially constant rate;

a sawtooth wave generator 'receiving the output of .said

current amplifier;

means `for triggering said sawtooth wave .generator at a predetermined frequency;

a pulse .generator connected =.to said -saw'tooth wave generator and .acting to produce a pulse whenever the amplitude of said sa-wtooth Wave reaches a predetermined reference level;

means for applying the output of said pulse generator to control the operation of said second gate;

a source of squarewave voltage of a constant frequency 'higher than that at which lsaid sawtoo'th Wave generator is triggered; land means for applying a square wave voltage from said source to control the loperation of said rst gate;

whereby, when both said gates are concurrently closed, current will flow into said capacitor from said source of replenishment energy =to compensate for a `diminution thereof and `a restoration Iof such energy to a level where it will be representative of the data .tobe remembered.

References Cited UNITED STATES PATENTS 3,014,169 12/1961 MacIntyre 320-1 3,050,673 8/1962` Widmer 320-41 3,171,986 3/1965 Bonner'et al 307-109 BERNARD KONIOK, Primary Examiner.

I. F. BREIMAYER, Assistant Examiner.

Claims (1)

1. 5. IN A COMPUTER INCORPORATING AN ENERGY-STORAGE DEVICE OF THE CAPACITANCE TYPE WHICH IS ADAPTED TO RECEIVE ENERGY REPRESENTATIVE OF DATA TO BE REMEMBERED, AND WHICH ENERGY IS SUBJECT TO GRADUAL DIMINUTION OVER A PERIOD OF TIME SUBSEQUENT TO ITS RECEPTION BY SAID DEVICE SO THAT AFTER SUCH PERIOD OF TIME SAID ENERGY IS NO LONGER REPRESENTATIVE OF THE DATA SUPPLIED THERETO, THE IMPROVEMENT WHICH COMPRISES A CIRCUIT FOR MAINTAINING A PREDETERMINED ENERGY LEVEL ON SAID CAPACITOR WHICH IS DETERMINED BY THE MAGNITUDE OF THE STORED ENEERGY AT ANY GIVEN INSTANT OF TIME, SAID CIRCUIT COMPRISING: A CAPACITOR ADAPTED TO RECEIVE ENERGY REPRESENTATIVE OF DATA TO BE REMEMBERED; A CURRENT AMPLIFIER IN SERIES WITH SAID CAPACITOR AND TO WHICH THE OUTPUT OF SAID CAPACITOR IS APPLIED; A FIRST GATE; A SECOND GATE; A SOURCE OF REPLENISHMENT ENERGY FOR SAID CAPACITOR; MEANS FOR ELECTRICALLY CONNECTING SAID TWO GATES AND SAID SOURCE OF REPLENISHMENT ENERGY FOR SAID CAPACITOR; ANOTHER, THE RESULTING SERIES COMBINATION BEING CONNECTED TO SAID CAPACITOR AND IN PARALLEL WITH SAID CURRENT AMPLIFIER; A DISCHARGE NETWORK CONNECTED IN PARALLEL WITH THE SERIES COMBINATION OF SAID TWO GATES AND SAID SOURCE OF REPLENISHMENT ENERGY, SAID DISCHARGE NETWORK ACTING TO DIMINISH THE CHARGE ON SAID CAPACITOR AT AN ESSENTIALLY CONSTANT RATE; A SAWTOOTH WAVE GENERATOR RECEIVING THE OUTPUT OF SAID CURRENT AMPLIFIER; MEANS FOR TRIGGERING SAID SAWTOOTH WAVE GENERATOR AT A PREDETERMINED FREQUENCY; A PULSE GENERATOR CONNECTED TO SAID SAWTOOTH WAVE GENERATOR AND ACTING TO PRODUCE A PULSE WHENEVER THE AMPLITUDE OF SAID SAWTOOTH WAVE REACHES A PREDETERMINED REFERENCE LEVEL; MEANS FOR APPLYING THE OUTPUT OF SAID PULSE GENERATOR TO CONTROL THE OPERATION OF SAID SECOND GATE; A SOURCE OF SQUARE-WAVE VOLTAGE OF A CONSTANT FREQUENCY HIGHER THAN THAT AT WHICH SAID SAWTOOTH WAVE GENERATOR IS TRIGGERED; AND MEANS FOR APPLYING A SQUARE WAVE VOLTAGE FROM SAID SOURCE TO CONTROL THE OPERATION OF SAID FIRST GATE; WHEREBY, WHEN BOTH SAID GATES ARE CONCURRENTLY CLOSED, CURRENT WILL FLOW INTO SAID CAPACITOR FROM SAID SOURCE OF REPLENISHMENT ENERGY TO COMPENSSATE FOR A DIMINUTION THEREOF AND A RESTORATION OF SUCH ENERGY TO A LEVEL WHERE IT WILL BE REPRESENTATIVE OF THE DATA TO BE REMEMBERED.

Images